ADOPTING ARTIFICIAL NEURAL NETWORK FOR WEAR INVESTIGATION OF BALL BEARING MATERIALS UNDER PURE SLIDING CONDITION

Abstract

In the industry, ball bearings are the most widely used machine element. The ball materials may differ in various bearing applications. Wear of the ball and recess after a period of use is the most common cause of ball bearing failure. The present study aims to develop the artificial neural network model for assessing the wear of different ball bearing materials. A wear test method has been followed as suggested by the ASTM-G99 standard. The pin on disc apparatus was selected to conduct numerous trials. L9 array was considered to design the experiments. The factors considered for this study were load, time, and sliding speed. Based on the results obtained, ANN code was proposed to evaluate wear using numerous test parameters. The results obtained from the proposed model are nearly similar to experimental results, which would be evidence for the correctness of the model. The proposed neural network model can be used in numerous applications with given parameters.

Keywords

• anova
• artificial neural network
• design of experiments wear analysis